O uso de localidade de referência para otimizar consultas em arquiteturas par-a-par

• Ano de defesa: 2003
• Autor(a) principal: Marcelo Werneck Barbosa
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: Ciência da computação; Computação; Cliente/servidor (Computação); Redes locais de computação; Redes de computadores
• Link de acesso: http://hdl.handle.net/1843/BUBD-9KLMPJ

Services on the Internet are evolving from centralized client-server architectures to fully distributed architectures. Systems based on such architectures are called peer-to-peer systems (P2P), and end-hosts participating in such systems are called peers. A fundamentalproblem that confronts peer-to-peer applications is to efficiently locate the node that stores a particular data item. The surging increase in the popularity of peer-to-peer applications has led to a dramatic need for a scalable and high performance content location protocol. Current search algorithms in peer-to-peer systems do not take each node's interest intoaccount. It is known that users tend to work and relate to each other in groups. A group of users, although not always located in geographical proximity, tends to use the same set of resources (files). We show that groups in peer-to-peer file sharing systems can be identified and search mechanisms in these systems should take the existence and characteristics of these groups into account. This work describes a distributed algorithm for peers to self-organize into clusters basedon interests (peer communities). Each peer maintains a community of peers which share similar interests. Content is located by querying peers in one's community. Peers that have content in common share the same interests. The concept of the algorithm is that there is greater chance to find a file one node is looking for in its own community than in the remainder of the network. Based on this principle, every time a node issues a query for a file, it will first check if it can be found within the community; otherwise, it will be searched in the other hosts. Summarizing, this work aims at implementing an algorithm that exploits locality in interests to efficiently allow content location and retrieval.